Integrated feature-scale simulation framework for atomic layer deposition incorporating transient transport and kinetics

Modeling atomic layer deposition (ALD) for realistic three-dimensional (3D) features under realistic process conditions is challenging because it requires accounting for the dynamic progression of precursor penetration and surface saturation. To address this challenge, we present a framework for transient feature-scale simulations of the ALD of Al2O3 using trimethylaluminum. This framework resolves the coupled evolution of precursor transport and surface reaction kinetics governing film growth in 3D space. Molecular transport within 3D holes is treated in the collision-less (Knudsen) flow regime, whereas surface adsorption is described using an extended Kisliuk-type sticking formulation that captures the coverage-dependent sticking behavior through a dimensionless parameter. The model captures time-dependent local flux redistribution and surface coverage evolution during a single ALD half-cycle, without relying on the steady-state and ideal saturation assumptions. Using this framework, we compute spatially and temporally resolved precursor flux distributions and evaluate the transient surface coverage profiles along the feature depth. Simulations reproduce the experimentally observed growth-per-cycle trends across varying precursor exposure times and reveal non-ideal conformality, with thickness gradients decreasing from ∼50% to ∼10% as the precursor exposure time increases from 0.2 to 3.0 s. By systematically varying key process parameters, including the sticking coefficient and pulse duration, we elucidate the role of multiple precursor re-emission in driving nonlinear growth dynamics at the bottom and sidewalls of the feature. The results also identify operating regimes that yield realistic conformality. Overall, the framework provides mechanistic insights and a predictive approach for optimizing the process conditions to improve conformality.